Operation and performance of the ATLAS Tile Calorimeter in Run 1

Abstract

The Tile Calorimeter is the hadron calorimeter covering the central region of the ATLAS experiment at the Large Hadron Collider. Approximately 10,000 photomultipliers collect light from scintillating tiles acting as the active material sandwiched between slabs of steel absorber. This paper gives an overview of the calorimeter's performance during the years 2008-2012 using cosmic-ray muon events and proton-proton collision data at centre-of-mass energies of 7 and 8 TeV with a total integrated luminosity of nearly 30 fb ${}^{-1}$ . The signal reconstruction methods, calibration systems as well as the detector operation status are presented. The energy and time calibration methods performed excellently, resulting in good stability of the calorimeter response under varying conditions during the LHC Run 1. Finally, the Tile Calorimeter response to isolated muons and hadrons as well as to jets from proton-proton collisions is presented. The results demonstrate excellent performance in accord with specifications mentioned in the Technical Design Report.

Fig. 1

The layout of the TileCal cells, denoted by a letter (A to E) plus an integer number. The A-layer is closest to the beam-line. The naming convention is repeated on each side of $\eta =0$

Fig. 2

The reference pulse shapes for high gain and low gain, shown in arbitrary units [6]

Fig. 3

The relative difference between the online channel energy ($E_{\mathrm{DSP}}$) calculated using the non-iterative OF method and the offline ($E_{\mathrm{OFLI}}$) channel energy reconstruction using the iterative OF method, as a function of the phase computed by the DSP ($t_{\mathrm{DSP}}$) with no correction (circles) and with application of the parabolic correction (squares) as a function of phase ($\tau$). The error bars are the standard deviations (RMS) of the relative difference distribution. Data are shown for collisions in 2011

Fig. 4

The absolute difference between the energies reconstructed using the optimal filtering reconstruction method with the non-iterative ($E_{\mathrm{OFLNI}}$) and iterative ($E_{\mathrm{OFLI}}$) signal reconstruction methods as a function of energy. The black markers represent mean values of $E_{\mathrm{OFLNI}}$–$E_{\mathrm{OFLI}}$ per a bin of $E_{\mathrm{OFLNI}}$. The parabolic correction is applied to $E_{\mathrm{OFLNI}}$. The data shown uses high $p_{\text{T}}$ ($>20$ GeV) isolated muons from $\sqrt{s}=7$ TeV collisions recorded in 2010

Fig. 5

Left: the mean cell reconstructed time (average of the times in the two channels associated with the given cell) as measured with jet events. The mean cell time decreases with the increase of the cell energy due to the reduction of the energy fraction of the slow hadronic component of hadronic showers [26, 27]. Right: example of the channel reconstructed time in jet events in 2011 data, with the channel energy between 2 and 4 GeV. The solid line represents the Gaussian fit to the data

Fig. 6

An example of timing jumps detected using the laser (full red circles) and physics (open black circles) events (left) before and (right) after the correction. The small offset of about 2 ns in collision data is caused by the energy dependence of the reconstructed time in jet events (see Fig. 5, left). In these plots, events with any energy are accepted to accumulate enough statistics

Fig. 7

Impact of the timing jump corrections on the reconstructed channel time in jets from collision data. Shown are all high-gain channels with $E_{\mathrm{ch}}>4$ GeV associated with a reconstructed jet. The plot represents $1.3{\mathrm{fb}}^{-1}$ of pp collision data acquired in 2012

Fig. 8

The $\varphi$-averaged electronic noise (RMS) as a function of $\eta$ of the cell, with both contributing read-out channels in high-gain mode. For each cell the average value over all modules is taken. The statistical uncertainties are smaller than the marker size. Values are extracted using all the calibration runs used for the 2011 data reprocessing. The different cell types are shown separately for each layer: A, BC, D, and E (gap/crack). The transition between the long and extended barrels can be seen in the range $0.7<|\eta |<1.0$

Fig. 9

Ratio of the RMS to the width ($\sigma$) of a single Gaussian fit to the electronic noise distribution for all channels averaged over 40 TileCal modules before (squares) and after (circles) the replacement of the LVPS. Higher-number channels are closer to the LVPS

Fig. 10

The TileCal cell energy spectrum at the electromagnetic (EM) scale measured in 2010 data. The distributions from collision data at 7 TeV, 2.36 TeV, and 0.9 TeV are superimposed with Pythia minimum-bias Monte Carlo and randomly triggered events

Fig. 11

The total noise per cell as a function of $|\eta |$ for $〈{\mu }_{\mathrm{run}}〉=15.7$, for the high–high gain combination. The data from a 2012 run, with a bunch spacing of 50 ns, are shown in black while the simulation is shown in blue. Four layers are displayed: layer A (top left), layer BC (top right), layer D (bottom left), and the special gap and crack cells (bottom right). The electronic noise component is shown in Fig. 8

Fig. 12

The area-normalised energy spectra in cells A12 over all TileCal modules for two different pile-up conditions $〈\mu 〉=20,30$ (left) and the total noise, computed as the standard deviation of the energy distribution in all A-layer cells, as a function of $〈\mu 〉$ (right) for data and minimum-bias MC simulation in 2012

Fig. 13

The signal paths for each of the three calibration systems used by the TileCal. The physics signal is denoted by the thick solid line and the path taken by each of the calibration systems is shown with dashed lines

Fig. 14

The plot on the left shows the average response (in arbitrary units, a.u.) from all cells within a given layer to the ${}^{137}$Cs source as a function of time from July 2009 to December 2012. The solid line represents the expected response, where the Cs source activity decreases in time by $-2.3\%$/year. The coloured band shows the declared precision of the Cs calibration ($\pm 0.3$%). The plot on the right shows the percentage difference of the response from the expectation as a function of time averaged over all cells in all partitions. Both plots display only the measurements performed with the magnetic field at its nominal value. The first points in the plot on the right deviate from zero, as the initial HV equalisation was done in June 2009 using Cs calibration data taken without the magnetic field (not shown in the plot). The increasing Cs response in the last three measurements corresponds to the period without collisions after the Run 1 data-taking finished

Fig. 15

The mean gain variation in the PMTs for each cell type averaged over $\varphi$ between a stand-alone laser calibration run taken on 21 April 2012 and a laser run taken before the collisions on 19 March 2012. For each cell type, the gain variation was defined as the mean of a Gaussian fit to the gain variations in the channels associated with this cell type. A total of 64 modules in $\varphi$ were used for each cell type, with the exclusion of known pathological channels

Fig. 16

Stability of the charge injection system constants for the low-gain ADCs (left) and high-gain ADCs (right) as a function of time in 2012. Values for the average over all channels and for one typical channel with the 0.7% systematic uncertainty are shown. Only good channels not suffering from damaged components relevant to the charge injection calibration are included in this figure

Fig. 17

The PMT current as measured by the slow integrator read-out as a function of cell $\eta$ and averaged over all modules for the three layers in the LB and EB, using minimum-bias data collected in 2011 at a fixed instantaneous luminosity ($1.9\times {10}^{32}$ cm${}^{-2}$s${}^{-1}$)

Fig. 18

The change of response seen in cell A13 by the minimum-bias, caesium, and laser systems throughout 2012. Minimum-bias data cover the period from the beginning of April to the beginning of December 2012. The Cs and laser results cover the period from mid-March to mid-December. The variation versus time for the response of the three systems was normalised to the first Cs scan (mid-March, before the start of collisions data-taking). The integrated luminosity is the total delivered during the proton–proton collision period of 2012. The down-drifts of the PMT gains (seen by the laser system) coincide with the collision periods, while up-drifts are observed during machine development periods. The drop in the response variation during the data-taking periods tends to decrease as the exposure of the PMTs increases. The variations observed by the minimum-bias and Cs systems are similar, both measurements being sensitive to PMT drift and scintillator irradiation

Fig. 19

One PMT of the EBC64 module with the largest gain variation. This plot presents a comparison between the gain expected from the HV instability (tiny dots), the one measured by the laser (open squares) and Cs (full circles) systems during the whole 2012 run. One HV point represents the averaged gain variation over one hour. The vertical structures are due to power cycles. There is very good agreement between the three methods, meaning that even large variations can be detected and handled by the TileCal monitoring and calibration systems

Fig. 20

The sources and amounts of integrated luminosity lost due to Tile Calorimeter data quality problems in 2012 as a function of time. The primary source of luminosity losses comes from the stop-less read-out link (ROL) removal in the extended barrels accounting for 45.2 pb${}^{-1}$ of this loss. Power cuts or trips of the 200V power supplies account for 22.6 pb${}^{-1}$. The last 4.9 pb${}^{-1}$ of losses stem from Laser Calibration ROD (LASTROD) busy events. The loss of 31.3 pb${}^{-1}$ due to a $-25$ ns timing shift in EBC are recovered after the data are reprocessed with updated timing constants. Each bin in the plot represents about two weeks of data-taking

Fig. 21

The percentage of the TileCal cells that are masked in the reconstruction as a function of time starting from June 2010. Periods of recovery correspond to times of hardware maintenance when the detector is accessible due to breaks in the accelerator schedule. Each super-drawer LB (EB) failure corresponds to 0.43% (0.35%) of masked cells. The total number of cells (including gap, crack, and minimum-bias trigger scintillators) is 5198. Approximately 2.9% of cells were masked in February 2013, at the end of the proton–lead data-taking period closing the Run 1 physics programme

Fig. 22

Distributions of the energy deposited by cosmic muons per unit of path length, $\mathrm{\Delta }E/\mathrm{\Delta }x$, in the two cells in the long barrel, A3 (left) and D2 (right) obtained using 2008 experimental (full points) and simulated (solid lines) data. The A3 (D2) cell in the long barrel covers the region $0.2<|\eta |<0.3$ ($0.3<|\eta |<0.5$) and is located in the innermost (outermost) calorimeter layer. The function curve overlaid on top of the experimental data is a Landau distribution convolved with a Gaussian distribution

Fig. 23

Visualisation of the TileCal in the (z, r) plane showing the results for the fit parameters of Eq. (6) for (top) the mean double ratio response $\widehat{\mu }$ and (bottom) the non-uniformity in $\varphi$ of the double ratio $\widehat{s}$. Shown for all cells with $|\eta |>0.1$, using the 2012 data and MC simulation

Fig. 24

Distributions of $⟨E/p⟩$ as a function of $\eta$ (top left), $\varphi$ (top right), p (bottom left) and $⟨\mu ⟩$ (bottom right) measured in 2012. Only statistical uncertainties are shown. The sources of systematic uncertainty are discussed in the text

Fig. 25

Electromagnetic scale jet energy fraction in the TileCal for jets with $p_{\text{T}}>140$ GeV in the long barrel ($|{\eta }_{\mathrm{jet}}|<0.8$) for layer A (left), layer BC (middle), and layer D (right). Error bars represent statistical uncertainties

Fig. 26

Electromagnetic scale jet energy fraction in the TileCal for jets with $p_{\text{T}}>140$ GeV in the extended barrel ($1.0<|{\eta }_{\mathrm{jet}}|<1.5$) for layer A (left), layer BC (middle), and layer D (right). Error bars represent statistical uncertainties

Fig. 27

Electromagnetic scale jet energy fraction in the TileCal for jets after combining the photon-plus-jet and inclusive jet samples for the long barrel (left) and extended barrel (right). Error bars represent statistical uncertainties. The simulation of the inclusive jets done with Pythia 8 (thick solid line) and Herwig++ [54] (dotted line) MC generators can be seen, while $\gamma$ + jets events were simulated only with Pythia 8 (thin solid line)

Fig. 28

Cell time resolution as a function of cell energy associated with jets and muons from 2011 collision data. The data are fit to Eq. (7), with the fit results for the three constants shown within the figure. The parameters $p_1$ and $p_2$ corresponds to energy expressed in GeV. Statistical errors are included on the points and on the fit parameters. The statistical uncertainties are smaller than the markers identifying the data points